Nobody knows if seti@home has enough crunching computer or not...


log in

Advanced search

Questions and Answers : Wish list : Nobody knows if seti@home has enough crunching computer or not...

Author Message
Profile DanTheMan
Send message
Joined: 2 Dec 00
Posts: 12
Credit: 6,555,720
RAC: 994
United States
Message 1099359 - Posted: 21 Apr 2011, 8:30:53 UTC

Please, under Server Status or wherever you feel appropriate, let us users know if we are wasting electricity by doing work units redundantly (beyond the need for verification of correct computation) or if the project needs more computing power to keep up with radio telescope data. I find it unlikely that you have "just enough" computing power and you've never had too much (wasting electricity), or too little (not able to keep up with radio data).

I have a feeling that you have far more computing power than you need because you have never sent out an email to everyone asking for more computing power, just emails asking for donations.

Thank you. I've been a big fan since 2001 when my electronics teacher told me about this project and I've been running it ever since .. (and folding@home too)
____________

OzzFan
Volunteer tester
Avatar
Send message
Joined: 9 Apr 02
Posts: 13662
Credit: 31,494,772
RAC: 11,185
United States
Message 1099380 - Posted: 21 Apr 2011, 12:02:29 UTC - in response to Message 1099359.
Last modified: 21 Apr 2011, 12:21:26 UTC

Actually, SETI@Home has sent out emails asking for people to come back to the project because they need more computing power. People who have a RAC greater than 1 do not get this email.


SETI@Home will never ask existing users for more crunching power because A) that would be rude to ask for more and would turn off quite a few people and B) would be beyond the project's original request that you only donate spare CPU cycles when you already have your computer powered on.

Many participants get so carried away with the race for more credits that they create more waste cycles by building extra machines to remain powered on 24/7, and the most enthusiastic load these machines up with CUDA cards. The power bills on the mega-crunchers can be quite high.

The problem is, when the project has server problems, these people are the most vocal because they're not earning credits when the server is down, and if their caches run dry, they're "wasting" their electric bill - even though SETI@Home never asked for them to do this. These same people load their caches up to the maximum of 20 days, which seriously pounds the server with work request once their caches do run dry after an extended server outage.


So, the answer is yes, they need more volunteers, and you can donate as much spare CPU time as you wish. And if the servers go down, join a backup project to keep your computers busy.
____________

Profile DanTheMan
Send message
Joined: 2 Dec 00
Posts: 12
Credit: 6,555,720
RAC: 994
United States
Message 1099618 - Posted: 22 Apr 2011, 10:09:03 UTC

So does this mean that they are having to store away work units/tapes that haven't been processed yet because there is not enough computing power?
I doubt it. I know that the project must be getting enough computing power for it's main project(seti), but Astropulse and other side projects? Not sure. Perhaps the extra computing power is going towards running Astropulse on old tapes?

It would be VERY helpful to get some figures on the server status page about this..


____________

Profile DanTheMan
Send message
Joined: 2 Dec 00
Posts: 12
Credit: 6,555,720
RAC: 994
United States
Message 1099619 - Posted: 22 Apr 2011, 10:10:41 UTC - in response to Message 1099380.

Yes i totally agree with you about older machines crunching. 5 years from now I know my gaming laptop will be ridiculously inefficient at doing math compared to new laptops. Intel will probably be at 11nm by then.
____________

Profile DanTheMan
Send message
Joined: 2 Dec 00
Posts: 12
Credit: 6,555,720
RAC: 994
United States
Message 1099621 - Posted: 22 Apr 2011, 10:16:48 UTC - in response to Message 1099380.

So, are the guys @ berkeley planning on running the Enhanced (more detailed analysis) seti@home client on all old tapes (like the original ones from the late 1990s on up, and compute Astropulse on all previous tapes as well? This I can see will take quite a bit of power to crunch through while keeping up with new data from Aricebo.
____________

Profile ignorance is no excuse
Avatar
Send message
Joined: 4 Oct 00
Posts: 9529
Credit: 44,433,321
RAC: 0
Korea, North
Message 1099637 - Posted: 22 Apr 2011, 13:40:04 UTC - in response to Message 1099618.
Last modified: 22 Apr 2011, 13:41:43 UTC

So does this mean that they are having to store away work units/tapes that haven't been processed yet because there is not enough computing power?
I doubt it. I know that the project must be getting enough computing power for it's main project(seti), but Astropulse and other side projects? Not sure. Perhaps the extra computing power is going towards running Astropulse on old tapes?

It would be VERY helpful to get some figures on the server status page about this..


Storing tapes because there's to much work. Bite your tongue. In the last couple of years they've actually turned up the sensitivity of their search because we were completing the tasks to fast and they were running out of work.

the only old tapes that haven't been processed are the ones that are so heavily hit by local radar that the WU's are useless in a search. We've floated the idae that they go back over the old tapes when we the new app is released. Perhaps running a modified v7 to read only the new search parameters on the old WU's. We haven't heard back yet so we have to assume its probably not a priority, yet

Astropulse is ran at the same time as the multibeam so theres no back log. when the tapes are processed you'll see that Astropulse WU's are generated at about 1 per 30 Multibeam WU's. SO there is no real backlog of either type of WU. Astropulse isn't considered a side project. It is wholey part of S@H.

We've floated the idea about running astropulse on old tapes but we haven't heard anything about that. Perhaps its not necessary or needed yet
____________
In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

End terrorism by building a school

OzzFan
Volunteer tester
Avatar
Send message
Joined: 9 Apr 02
Posts: 13662
Credit: 31,494,772
RAC: 11,185
United States
Message 1099689 - Posted: 22 Apr 2011, 16:52:40 UTC - in response to Message 1099618.

So does this mean that they are having to store away work units/tapes that haven't been processed yet because there is not enough computing power?


Last I heard from Matt Lebofsky, yes. They don't actually store things on "tape" anymore, but on an offsite storage location that they can store and pull data from right over the internet.

I know that the project must be getting enough computing power for it's main project(seti), but Astropulse and other side projects? Not sure. Perhaps the extra computing power is going towards running Astropulse on old tapes?


Besides not having enough power for the main application "SETI@Home Enhanced Multibeam", the AstroPulse application does not get left-over computing power unless the user selects to run AP in their preferences.

I'll PM Matt to see if he can offer better answers to your questions.
____________

Profile DanTheMan
Send message
Joined: 2 Dec 00
Posts: 12
Credit: 6,555,720
RAC: 994
United States
Message 1099797 - Posted: 22 Apr 2011, 21:30:26 UTC

Wow so Astropulse is not keeping up with new data from Arecibo and seti@home Enhanced hasn't been run on tapes from the late '90s thru 2006? Since seti@home Enhanced started in May of 2006 (wikipedia.

It would be awesome to get some details
____________

Questions and Answers : Wish list : Nobody knows if seti@home has enough crunching computer or not...

Copyright © 2014 University of California